1 //===--- ObjectFilePCHContainerOperations.cpp -----------------------------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 
10 #include "clang/CodeGen/ObjectFilePCHContainerOperations.h"
11 #include "CGDebugInfo.h"
12 #include "CodeGenModule.h"
13 #include "clang/AST/ASTContext.h"
14 #include "clang/AST/DeclObjC.h"
15 #include "clang/AST/Expr.h"
16 #include "clang/AST/RecursiveASTVisitor.h"
17 #include "clang/Basic/Diagnostic.h"
18 #include "clang/Basic/TargetInfo.h"
19 #include "clang/CodeGen/BackendUtil.h"
20 #include "clang/Frontend/CodeGenOptions.h"
21 #include "clang/Frontend/CompilerInstance.h"
22 #include "clang/Lex/HeaderSearch.h"
23 #include "clang/Lex/Preprocessor.h"
24 #include "clang/Serialization/ASTWriter.h"
25 #include "llvm/ADT/StringRef.h"
26 #include "llvm/Bitcode/BitstreamReader.h"
27 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
28 #include "llvm/IR/Constants.h"
29 #include "llvm/IR/DataLayout.h"
30 #include "llvm/IR/LLVMContext.h"
31 #include "llvm/IR/Module.h"
32 #include "llvm/Object/COFF.h"
33 #include "llvm/Object/ObjectFile.h"
34 #include "llvm/Support/Path.h"
35 #include "llvm/Support/TargetRegistry.h"
36 #include <memory>
37 #include <utility>
38 
39 using namespace clang;
40 
41 #define DEBUG_TYPE "pchcontainer"
42 
43 namespace {
44 class PCHContainerGenerator : public ASTConsumer {
45   DiagnosticsEngine &Diags;
46   const std::string MainFileName;
47   const std::string OutputFileName;
48   ASTContext *Ctx;
49   ModuleMap &MMap;
50   const HeaderSearchOptions &HeaderSearchOpts;
51   const PreprocessorOptions &PreprocessorOpts;
52   CodeGenOptions CodeGenOpts;
53   const TargetOptions TargetOpts;
54   const LangOptions LangOpts;
55   std::unique_ptr<llvm::LLVMContext> VMContext;
56   std::unique_ptr<llvm::Module> M;
57   std::unique_ptr<CodeGen::CodeGenModule> Builder;
58   std::unique_ptr<raw_pwrite_stream> OS;
59   std::shared_ptr<PCHBuffer> Buffer;
60 
61   /// Visit every type and emit debug info for it.
62   struct DebugTypeVisitor : public RecursiveASTVisitor<DebugTypeVisitor> {
63     clang::CodeGen::CGDebugInfo &DI;
64     ASTContext &Ctx;
65     DebugTypeVisitor(clang::CodeGen::CGDebugInfo &DI, ASTContext &Ctx)
66         : DI(DI), Ctx(Ctx) {}
67 
68     /// Determine whether this type can be represented in DWARF.
69     static bool CanRepresent(const Type *Ty) {
70       return !Ty->isDependentType() && !Ty->isUndeducedType();
71     }
72 
73     bool VisitImportDecl(ImportDecl *D) {
74       if (!D->getImportedOwningModule())
75         DI.EmitImportDecl(*D);
76       return true;
77     }
78 
79     bool VisitTypeDecl(TypeDecl *D) {
80       // TagDecls may be deferred until after all decls have been merged and we
81       // know the complete type. Pure forward declarations will be skipped, but
82       // they don't need to be emitted into the module anyway.
83       if (auto *TD = dyn_cast<TagDecl>(D))
84         if (!TD->isCompleteDefinition())
85           return true;
86 
87       QualType QualTy = Ctx.getTypeDeclType(D);
88       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
89         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
90       return true;
91     }
92 
93     bool VisitObjCInterfaceDecl(ObjCInterfaceDecl *D) {
94       QualType QualTy(D->getTypeForDecl(), 0);
95       if (!QualTy.isNull() && CanRepresent(QualTy.getTypePtr()))
96         DI.getOrCreateStandaloneType(QualTy, D->getLocation());
97       return true;
98     }
99 
100     bool VisitFunctionDecl(FunctionDecl *D) {
101       if (isa<CXXMethodDecl>(D))
102         // This is not yet supported. Constructing the `this' argument
103         // mandates a CodeGenFunction.
104         return true;
105 
106       SmallVector<QualType, 16> ArgTypes;
107       for (auto i : D->parameters())
108         ArgTypes.push_back(i->getType());
109       QualType RetTy = D->getReturnType();
110       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
111                                           FunctionProtoType::ExtProtoInfo());
112       if (CanRepresent(FnTy.getTypePtr()))
113         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
114       return true;
115     }
116 
117     bool VisitObjCMethodDecl(ObjCMethodDecl *D) {
118       if (!D->getClassInterface())
119         return true;
120 
121       bool selfIsPseudoStrong, selfIsConsumed;
122       SmallVector<QualType, 16> ArgTypes;
123       ArgTypes.push_back(D->getSelfType(Ctx, D->getClassInterface(),
124                                         selfIsPseudoStrong, selfIsConsumed));
125       ArgTypes.push_back(Ctx.getObjCSelType());
126       for (auto i : D->parameters())
127         ArgTypes.push_back(i->getType());
128       QualType RetTy = D->getReturnType();
129       QualType FnTy = Ctx.getFunctionType(RetTy, ArgTypes,
130                                           FunctionProtoType::ExtProtoInfo());
131       if (CanRepresent(FnTy.getTypePtr()))
132         DI.EmitFunctionDecl(D, D->getLocation(), FnTy);
133       return true;
134     }
135   };
136 
137 public:
138   PCHContainerGenerator(CompilerInstance &CI, const std::string &MainFileName,
139                         const std::string &OutputFileName,
140                         std::unique_ptr<raw_pwrite_stream> OS,
141                         std::shared_ptr<PCHBuffer> Buffer)
142       : Diags(CI.getDiagnostics()), MainFileName(MainFileName),
143         OutputFileName(OutputFileName), Ctx(nullptr),
144         MMap(CI.getPreprocessor().getHeaderSearchInfo().getModuleMap()),
145         HeaderSearchOpts(CI.getHeaderSearchOpts()),
146         PreprocessorOpts(CI.getPreprocessorOpts()),
147         TargetOpts(CI.getTargetOpts()), LangOpts(CI.getLangOpts()),
148         OS(std::move(OS)), Buffer(std::move(Buffer)) {
149     // The debug info output isn't affected by CodeModel and
150     // ThreadModel, but the backend expects them to be nonempty.
151     CodeGenOpts.CodeModel = "default";
152     CodeGenOpts.ThreadModel = "single";
153     CodeGenOpts.DebugTypeExtRefs = true;
154     // When building a module MainFileName is the name of the modulemap file.
155     CodeGenOpts.MainFileName =
156         LangOpts.CurrentModule.empty() ? MainFileName : LangOpts.CurrentModule;
157     CodeGenOpts.setDebugInfo(codegenoptions::FullDebugInfo);
158     CodeGenOpts.setDebuggerTuning(CI.getCodeGenOpts().getDebuggerTuning());
159     CodeGenOpts.DebugPrefixMap =
160         CI.getInvocation().getCodeGenOpts().DebugPrefixMap;
161   }
162 
163   ~PCHContainerGenerator() override = default;
164 
165   void Initialize(ASTContext &Context) override {
166     assert(!Ctx && "initialized multiple times");
167 
168     Ctx = &Context;
169     VMContext.reset(new llvm::LLVMContext());
170     M.reset(new llvm::Module(MainFileName, *VMContext));
171     M->setDataLayout(Ctx->getTargetInfo().getDataLayout());
172     Builder.reset(new CodeGen::CodeGenModule(
173         *Ctx, HeaderSearchOpts, PreprocessorOpts, CodeGenOpts, *M, Diags));
174 
175     // Prepare CGDebugInfo to emit debug info for a clang module.
176     auto *DI = Builder->getModuleDebugInfo();
177     StringRef ModuleName = llvm::sys::path::filename(MainFileName);
178     DI->setPCHDescriptor({ModuleName, "", OutputFileName,
179                           ASTFileSignature{{{~0U, ~0U, ~0U, ~0U, ~1U}}}});
180     DI->setModuleMap(MMap);
181   }
182 
183   bool HandleTopLevelDecl(DeclGroupRef D) override {
184     if (Diags.hasErrorOccurred())
185       return true;
186 
187     // Collect debug info for all decls in this group.
188     for (auto *I : D)
189       if (!I->isFromASTFile()) {
190         DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
191         DTV.TraverseDecl(I);
192       }
193     return true;
194   }
195 
196   void HandleTopLevelDeclInObjCContainer(DeclGroupRef D) override {
197     HandleTopLevelDecl(D);
198   }
199 
200   void HandleTagDeclDefinition(TagDecl *D) override {
201     if (Diags.hasErrorOccurred())
202       return;
203 
204     if (D->isFromASTFile())
205       return;
206 
207     // Anonymous tag decls are deferred until we are building their declcontext.
208     if (D->getName().empty())
209       return;
210 
211     // Defer tag decls until their declcontext is complete.
212     auto *DeclCtx = D->getDeclContext();
213     while (DeclCtx) {
214       if (auto *D = dyn_cast<TagDecl>(DeclCtx))
215         if (!D->isCompleteDefinition())
216           return;
217       DeclCtx = DeclCtx->getParent();
218     }
219 
220     DebugTypeVisitor DTV(*Builder->getModuleDebugInfo(), *Ctx);
221     DTV.TraverseDecl(D);
222     Builder->UpdateCompletedType(D);
223   }
224 
225   void HandleTagDeclRequiredDefinition(const TagDecl *D) override {
226     if (Diags.hasErrorOccurred())
227       return;
228 
229     if (const RecordDecl *RD = dyn_cast<RecordDecl>(D))
230       Builder->getModuleDebugInfo()->completeRequiredType(RD);
231   }
232 
233   void HandleImplicitImportDecl(ImportDecl *D) override {
234     if (!D->getImportedOwningModule())
235       Builder->getModuleDebugInfo()->EmitImportDecl(*D);
236   }
237 
238   /// Emit a container holding the serialized AST.
239   void HandleTranslationUnit(ASTContext &Ctx) override {
240     assert(M && VMContext && Builder);
241     // Delete these on function exit.
242     std::unique_ptr<llvm::LLVMContext> VMContext = std::move(this->VMContext);
243     std::unique_ptr<llvm::Module> M = std::move(this->M);
244     std::unique_ptr<CodeGen::CodeGenModule> Builder = std::move(this->Builder);
245 
246     if (Diags.hasErrorOccurred())
247       return;
248 
249     M->setTargetTriple(Ctx.getTargetInfo().getTriple().getTriple());
250     M->setDataLayout(Ctx.getTargetInfo().getDataLayout());
251 
252     // PCH files don't have a signature field in the control block,
253     // but LLVM detects DWO CUs by looking for a non-zero DWO id.
254     // We use the lower 64 bits for debug info.
255     uint64_t Signature =
256         Buffer->Signature
257             ? (uint64_t)Buffer->Signature[1] << 32 | Buffer->Signature[0]
258             : ~1ULL;
259     Builder->getModuleDebugInfo()->setDwoId(Signature);
260 
261     // Finalize the Builder.
262     if (Builder)
263       Builder->Release();
264 
265     // Ensure the target exists.
266     std::string Error;
267     auto Triple = Ctx.getTargetInfo().getTriple();
268     if (!llvm::TargetRegistry::lookupTarget(Triple.getTriple(), Error))
269       llvm::report_fatal_error(Error);
270 
271     // Emit the serialized Clang AST into its own section.
272     assert(Buffer->IsComplete && "serialization did not complete");
273     auto &SerializedAST = Buffer->Data;
274     auto Size = SerializedAST.size();
275     auto Int8Ty = llvm::Type::getInt8Ty(*VMContext);
276     auto *Ty = llvm::ArrayType::get(Int8Ty, Size);
277     auto *Data = llvm::ConstantDataArray::getString(
278         *VMContext, StringRef(SerializedAST.data(), Size),
279         /*AddNull=*/false);
280     auto *ASTSym = new llvm::GlobalVariable(
281         *M, Ty, /*constant*/ true, llvm::GlobalVariable::InternalLinkage, Data,
282         "__clang_ast");
283     // The on-disk hashtable needs to be aligned.
284     ASTSym->setAlignment(8);
285 
286     // Mach-O also needs a segment name.
287     if (Triple.isOSBinFormatMachO())
288       ASTSym->setSection("__CLANG,__clangast");
289     // COFF has an eight character length limit.
290     else if (Triple.isOSBinFormatCOFF())
291       ASTSym->setSection("clangast");
292     else
293       ASTSym->setSection("__clangast");
294 
295     LLVM_DEBUG({
296       // Print the IR for the PCH container to the debug output.
297       llvm::SmallString<0> Buffer;
298       clang::EmitBackendOutput(
299           Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts, LangOpts,
300           Ctx.getTargetInfo().getDataLayout(), M.get(),
301           BackendAction::Backend_EmitLL,
302           llvm::make_unique<llvm::raw_svector_ostream>(Buffer));
303       llvm::dbgs() << Buffer;
304     });
305 
306     // Use the LLVM backend to emit the pch container.
307     clang::EmitBackendOutput(Diags, HeaderSearchOpts, CodeGenOpts, TargetOpts,
308                              LangOpts, Ctx.getTargetInfo().getDataLayout(),
309                              M.get(), BackendAction::Backend_EmitObj,
310                              std::move(OS));
311 
312     // Free the memory for the temporary buffer.
313     llvm::SmallVector<char, 0> Empty;
314     SerializedAST = std::move(Empty);
315   }
316 };
317 
318 } // anonymous namespace
319 
320 std::unique_ptr<ASTConsumer>
321 ObjectFilePCHContainerWriter::CreatePCHContainerGenerator(
322     CompilerInstance &CI, const std::string &MainFileName,
323     const std::string &OutputFileName,
324     std::unique_ptr<llvm::raw_pwrite_stream> OS,
325     std::shared_ptr<PCHBuffer> Buffer) const {
326   return llvm::make_unique<PCHContainerGenerator>(
327       CI, MainFileName, OutputFileName, std::move(OS), Buffer);
328 }
329 
330 StringRef
331 ObjectFilePCHContainerReader::ExtractPCH(llvm::MemoryBufferRef Buffer) const {
332   StringRef PCH;
333   auto OFOrErr = llvm::object::ObjectFile::createObjectFile(Buffer);
334   if (OFOrErr) {
335     auto &OF = OFOrErr.get();
336     bool IsCOFF = isa<llvm::object::COFFObjectFile>(*OF);
337     // Find the clang AST section in the container.
338     for (auto &Section : OF->sections()) {
339       StringRef Name;
340       Section.getName(Name);
341       if ((!IsCOFF && Name == "__clangast") || (IsCOFF && Name == "clangast")) {
342         Section.getContents(PCH);
343         return PCH;
344       }
345     }
346   }
347   handleAllErrors(OFOrErr.takeError(), [&](const llvm::ErrorInfoBase &EIB) {
348     if (EIB.convertToErrorCode() ==
349         llvm::object::object_error::invalid_file_type)
350       // As a fallback, treat the buffer as a raw AST.
351       PCH = Buffer.getBuffer();
352     else
353       EIB.log(llvm::errs());
354   });
355   return PCH;
356 }
357